摘要
针对矿山旋回式破碎机结构庞大,运行机理复杂以及故障关联性较强的问题,在利用传感器系统采集多源异质数据的基础上,提出一种基于支持向量机(SVM)和D-S证据理论的破碎机故障诊断方法。首先,利用所采集的破碎机的振动、声音、温度和压力数据信息构建多个特征证据体;然后,使用"一对一"多分类SVM对每个证据体进行初步的训练、测试、分析诊断;其次,利用D-S证据理论将初步的SVM诊断结果进行融合,得出最终结果;结果表明:D-S融合后的故障诊断正确率平均为93.2%,与融合前的单一证据体SVM故障诊断正确率高16.8个百分点。由此可得,基于SVM和D-S证据理论的矿山破碎机故障智能诊断方法准确、可靠,在矿山企业具有较高的应用实践价值。
Aiming at the huge structure,complex operation mechanism and strong faults correlation in mine gyratory crusher,the multi-source heterogeneous data was collected by sensor system.A fault diagnosis method of crusher based on Support Vector Machine（SVM）and D-S evidence theory was put forward.Firstly,aplurality of characteristic evidences bodys was constructed based on the acquired vibration,sound,temperature and pressure data of the crusher.Then,the ＂one-to-one＂ multiclassification SVM was used to conduct preliminary training,testing and analysis of each evidence body.Secondly,the D-S evidence theory was used to fuse the preliminary SVM diagnosis results to get the final result.The results showed that the average accuracy of fault diagnosis after the D-S fusion was 93.2%,and 16.8 percentage points higher than that of the single evidence SVM fault diagnosis before fusion.It can be concluded that the intelligent fault diagnosis method of mine crusher based on SVM and D-S evidence theory was accurate and reliable and had a high practical value in mining enterprises.
作者
王甜甜
卢才武
李发本
WANG Tiantian;LU Caiwu;LI Faben(Xi＇an University of Architecture and Technology, Xi＇an, Shaanxi 710055,China;China Molybdenum Co., Ltd, Luoyang, Henan 471000, China)
出处
《矿业研究与开发》
CAS
北大核心
2018年第5期69-73,共5页
Mining Research and Development
基金
陕西省自然科学基金项目(2017JM7005)
